HomeAIAI ModelsTencent Adds Hy4 Flash to TokenHub After the Hy4 Launch

Tencent Adds Hy4 Flash to TokenHub After the Hy4 Launch

Tencent has added Hy4 flash to its TokenHub model catalogue. A TokenHub product-updates page dated 16 November 2026 lists “新增支持 Hy4 flash 模型” — newly supported Hy4 flash. The TokenHub language-model API overview, last updated at 12:15 on 16 November 2026, gives the model parameter as hy4-flash and marks it as compatible with the OpenAI Chat Completions and Responses APIs and the Anthropic Messages API, the same three protocols already listed for Hy4.

The flash variant arrives after Tencent’s Hy4 launch, which TSN covered when the full model opened with vision and computer use, and with weights still to follow. Official Tencent documentation confirms that Hy4 flash is now a TokenHub model ID. The product-updates and API-overview pages stop short of restating parameter counts, token prices or test scores.

Specs, prices and test scores are from TechNode’s reporting

TechNode reported that Hy4 flash has 224 billion parameters and activates 12 billion per forward pass, supports multimodal inputs and a context window of up to one million tokens, and has API list prices of $0.07 per million input tokens, $0.281 per million output tokens and $0.003 per million cached input tokens. Those figures appear in TechNode’s article, not in the TokenHub updates or API-overview pages cited above.

The same TechNode report said that, on Tencent’s internal WorkIndex benchmark, Hy4 flash cost one-twentieth as much per task as the full Hy4 model. In a separate blind test, TechNode wrote, 85 WorkBuddy users rated outputs across 259 tasks, and Tencent said the model scored close to Claude Haiku 5.5 in user satisfaction and had the lowest cost and shortest completion time among the models tested. That comparison is Tencent’s claim as reported by TechNode.

TechNode also said the model is available in WorkBuddy and Yuanbao as well as through TokenHub, that WorkBuddy users can use it for free through 30 November, and that TokenHub is offering API access at half the list price. Tencent, it reported, plans to release the model’s weights and make it available on OpenRouter. No date is given for those two steps, and they are not confirmed on the official TokenHub pages used here.

The WorkBuddy offer sits on a product Tencent has already been pushing as an enterprise assistant. TSN previously reported Tencent’s claim that WorkBuddy had passed 9 billion tasks as it previewed the Hy4 release. A wider set of Tencent developer tools, including the open-sourced TeamAI harness for coding agents, is a separate track from this flash listing.

A catalogue entry is not a published official price list

The official fact on 16 November is narrow and solid: TokenHub now lists Hy4 flash, under the model ID hy4-flash, on the same OpenAI- and Anthropic-compatible API as Hy4. That is enough to treat the launch as real. It is not enough to treat the parameter counts, the dollar prices, the WorkIndex “one-twentieth” claim or the Haiku comparison as figures Tencent has posted on those documentation pages.

TechNode’s reporting may be accurate. It may also be repeating a Tencent briefing that has not yet been written into the public docs. Until Tencent publishes those numbers on TokenHub, they should be read as TechNode’s account of Tencent’s claims: a 224-billion-parameter model with 12 billion active, a one-million-token window, list prices of $0.07 / $0.281 / $0.003 per million tokens, a steep cut in per-task cost on an internal benchmark, and a favourable showing in a Tencent-run blind test. The free WorkBuddy window through 30 November and the half-price TokenHub API offer belong in the same attributed box, as do the planned weight release and OpenRouter listing.

The earlier Hy4 launch already left weights as a promise rather than a delivery. Flash, as reported, repeats that pattern. Anyone costing a workload should wait for an official TokenHub price table, or treat the TechNode list prices as unverified against the docs, before assuming a 95 per cent saving or a standing discount.

Sources

Related stories

Follow @tsnmediahub on X for breaking AI news as it happens.

Follow on X

Share this story

Latest stories

More in this category

Latest stories

Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.